Coordinates of the key facial feature points can be captured by motion capture system OPTOTRAK with real-time character and high accuracy. The facial model is considered as an undirected weighted graph. By iteratively subdividing the related triangle edges, the geodesic distance between points on the model surface is finally obtained. The RBF （Radial Basis Functions） interpolation technique based on geodesic distance is applied to generate deformation of the facial mesh model. Experimental results demonstrate that the geodesic distance can explore the complex topology of human face models perfectly and the method can generate realistic facial expressions.
A hand gesture recognition method is presented for human-computer interaction, which is based on fingertip localization. First, hand gesture is segmented from the background based on skin color characteristics. Second, feature vectors are selected with equal intervals on the boundary of the gesture, and then gestures＇ length normalization is accomplished. Third, the fingertip positions are determined by the feature vectors＇ parameters, and angles of feature vectors are normalized. Finally the gestures are classified by support vector machine. The experimental results demonstrate that the proposed method can recognize 9 gestures with an accuracy of 94.1%.
提出了一种对面部遮挡具有鲁棒性的表情识别方法.首先,基于鲁棒主成分分析(RPCA,Robust Principal Component Analysis)对待识别人脸进行重构,并对重构人脸和待识别人脸的差值图像进行显著性检测得到面部遮挡区域;其次,将待识别人脸的遮挡区域由RPCA重构人脸的相应区域进行替换,并由权值更新的AdaBoost分类器对遮挡区域重构后的人脸进行表情识别.在BHU(Beihang University)人脸表情数据库和日本女性表情数据库上进行了各种遮挡情况下的表情识别实验,获得了比AdaBoost方法更好的识别结果,说明基于RPCA和AdaBoost的表情识别方法对多种面部遮挡具有较强的鲁棒性.
